Art Original

Indonesian Journal of Electrical Engineering and Informatics (IJEEI) Vol 14, No 2, June 2026 | Deepfake Detection Based on Deep Learning and Quantum Particle Swarm Optimization

---

Recently, many fake images and videos can be created and manipulated easily by employing Artificial Intelligence technologies, mainly based on Generative Adversarial Network (GAN). These synthesized media (images or videos) can be used later in unethical ways to create panic among individuals. This type of technology, which synthesizes human images and videos, is called deepfake. This paper proposes a deepfake detection methodology relying on Deep\\\\\\\\r\\\\\\\\nLearning (DL) approach and quantum particle swarm optimization (QPSO). The input image undergoes preprocessing by choosing the Region of Interest (ROI), followed by the application of an image downsampling algorithm to minimize the dimension of the new sub-image and thus decrease processing time costs. After that, the image has been normalized. The optimal subset of dataset has been determined by adapting QPSO during training phase. To perform classification of images and videos, ResNet50 model accomplishes this task perfectly. Experiments indicate that the suggested algorithm demonstrates a significant level of accuracy on ”Real and Fake Face Detection ” and ” Celeb-DF” datasets. The proposed approach achieves accuracy 99.6 % on Celeb-DF and 99.8 % on Real and Fake Face Detection dataset.

---

Ketersediaan

Informasi Detail

Judul Seri

-

No. Panggil

050

Penerbit

: Indonesian Journal of Electrical Engineering and Informatics (IJEEI)., 2026

Deskripsi Fisik

-

Bahasa

English

ISBN/ISSN

2089-3272

Klasifikasi

050

Tipe Isi

-

Tipe Media

-

Tipe Pembawa

-

Edisi

-

Subjek

JURNAL

Info Detail Spesifik

-

Pernyataan Tanggungjawab

-

Versi lain/terkait

Lampiran Berkas

Komentar

---

Anda harus masuk sebelum memberikan komentar